1. Field of the Invention
The present invention relates to a voice generator for generating voice corresponding to a text item and, particularly, to a multilingual processing technique for the voice generator.
2. Description of the Related Art
Japanese Unexamined Patent Application Publication No. 8-30287 or 8-95597 discloses a text-to-speech (TTS) engine, which is known as a technique for voice generators to generate voice corresponding to text items.
To process multiple languages, the TTS engine stores processes and data for the conversion of a text item to voice, which are prepared for each language, and switches the processes and data according to the language of the text item, thereby generating voice corresponding to the language.
Unfortunately, the conversion with the use of the processes and data for each language requires a larger-scale resource and a complicated process.